The introduction of a social gradient in mortality in the Destinie 2 model

We use here the mortality tables by education level recently published by Blanpain (2016b) to significantly improve the projected differential mortality in the Destinie 2 model that was usually only based on Insee mortality projections broken down by age and sex categories. We show that the relational proportionality method better predicts the evolution of differential mortality over the recent past than the Brass relationality method, used in other microsimulation models. The introduction of heterogeneous mortality according to education level aims at better reproducing the positive correlation that exists between life expectancy and pension levels. To quantify it, we compare the elasticities of survival to the amount of the retirement pension, by sex and age, obtained by microsimulation with those estimated on the Échantillon Interrégimes de Retraités (EIR). The simulated elasticities are lower than the empirical elasticities, especially for women. For men, however, they are statistically significant and at least two-thirds of the empirically observed values. At the aggregate level, this introduction also slightly increases projected pension expenditure.